Land use change has caused natural ecosystems to become more fragmented.  This has significant, largely negative implications for their native biodiversity, through the effects on species of area reduction, edge exposure and isolation of remaining ecosystem patches, as well as through interruption of ecosystem processes and associated ecosystem degradation.  Ecosystem fragmentation and its ecological effects are therefore significant drivers of biodiversity loss. Forests are the ecosystem type for which measurement of spatial changes in extent and configuration is most feasible and for which there is the greatest body of research to aid in interpreting the significance of changes.

At present, changes in forest fragmentation through time are poorly understood because we do not have meaningful ways to quantify those changes. In developing this indicator, we are analysing existing datasets on species composition and abundance in fragmented forest landscapes to determine which measures of fragmentation are most closely related to the biological changes seen in the field.

This ‘meta-analysis’ currently incorporates data on 1-5 taxa (e.g. understorey plants, trees, birds, bats, reptiles, small mammals, beetles) from more than a dozen fragmented forest landscapes around the world. For each landscape and taxon, the variation in an index of community similarity for each forest plot potentially affected by fragmentation is assessed in relation to measures of fragmentation. The resulting statistical relations are used to predict community composition for every forest pixel on land-cover maps of the study areas, which are used to calculate a landscape-level estimate of compositional change, termed ‘BioFrag’.

This approach has the advantage of incorporating spatial patterns of observed species responses to fragmentation operating at multiple spatial scales.  Performing similar analyses for 20-50 different datasets from around the world will help to select the best possible form of BioFrag for application at regional and global scales.

This indicator is derived from remotely sensed data on forest cover and is therefore potentially applicable at scales ranging from individual landscapes to continental and global.

For any given landscape, such as that shown below for an Atlantic forest landscape in Brazil, the fragmentation index values for each pixel can be used tocalculate a single summary statistic of biodiversity change based on fragmentation (‘Biofrag’).

BioFrag: 0.794.

Change in that statistic over time could be tracked in a simple time servies graph.
Source: UNEP-WCMC 

Forest fragmentation is a significant driver of biodiversity loss. The fully developed and implemented indicator will enable users to identify specific changes in forest configuration that may be associated with biodiversity loss.  Ultimately, it will be possible to detect changes in trends as well as to identify locations where such changes are large.

Based on these trends and spatial information, decision-makers may wish to adjust land use policies to discourage changes that result in further forest fragmentation.

The meta-analysis will be expanded to include as many additional data sets as possible.  We welcome suggestions of additional field studies that assess species composition at accurately located points within fragmented forest landscapes and include associated land cover data. The indicator selected will be applied at continental scale.

The following papers have been submitted for publication and accepted subject to revision:

Lafortezza, R., Coomes, D., Kapos, V. and  Ewers, R. Assessing the impacts of forest fragmentation in New Zealand: scaling from survey plots to landscapes. Global Ecology and Biogeography.

Ewers, R.M., Kapos, V., Coomes, D., Lafortezza, R. and Didham, R.K. Mapping community change in modified landscapes. Biological Conservation.